Dear Experts, I decided to open this question based on the requirement of suggestions or recommendations. If you have had related experiences with this question, please share the information with me. As I have known a little bit the RNN/LSMT network outperforms ANN.
Thanks a lot in advance,
Phimmasone
Generally no model is always better than another. It depends on the circumstances. In some circumstances, one model might be better. In some other circumstances, the other one might do better. Simulation studies might be necessary to uncover the circumstances necessary for a model to be the better model.
I agree with Ette Etuk . In order to implement simulation studies, you can use forecast evaluation workflows proposed in this paper:
Article Assessing Point Forecast Bias Across Multiple Time Series: M...
More specifically, the FEW-L2 workflow should be applicable, but it depends on how you optimize forecasts, please see relevant sections of the above paper for details and more references.
Andrey Davydenko and Vadym Slyusar , thank you very much for your information. It is very helpful for me.
- Badges
- Science topic